MuTect2 missing some variants called by MuTect

david.knoffdavid.knoff Boston, MAMember

Hello,

I was recently doing some QC on my MuTect2 variant analysis and noticed that in one of my samples a specific mutation was not called (An IDH1 R132H mutation). In this particular sample, the original MuTect-1.1.7 analysis did call this mutation. We have verified this mutation by several methods including IHC and visualizing the BAM file in IGV. Do you know why this could be occurring? Are there different hard filters for MuTect2 compared to MuTect-1.1.7?

Thank you for your feedback!
David

MuTect-1.1.7 script:
/path/to/java -Xmx2g -jar /apps/mutect/target/mutect-1.1.7.jar -T MuTect -R /path/to/human_g1k_v37.fasta --input_file:normal /path/to/CEPH_1347-02_L_001078_000001.dedup.cleaned.bam --input_file:tumor /path/to/BTXXX.dedup.cleaned.bam --normal_panel /path/to/150323_wesv2_mutect_pon.vcf --dbsnp /path/to/dbsnp_138.b37.vcf --cosmic /path/to/CosmicCodingMuts_v68_sorted.vcf -L /path/to/Agilent_exome_v2_44mb_1kGenomeB37_baits.interval_list -o /path/to/BTXXX_call_stats.out --coverage_file /path/to/BTXXX_coverage.wig.txt -vcf /path/to/BTXXX_WES_MuTect.vcf

MuTect2 script:
/path/to/java -Xmx2g -jar /apps/gatk3.5/target/executable/GenomeAnalysisTK.jar -T MuTect2 -R /path/to/human_g1k_v37.fasta -I:normal /path/to/CEPH_1347-02_L_001078_000001.dedup.cleaned.bam -I:tumor /path/to/BTXXX.dedup.cleaned.bam --normal_panel /path/to/150323_wesv2_mutect_pon.vcf --dbsnp /path/to/dbsnp_138.b37.vcf --cosmic /path/to/CosmicCodingMuts_v68_sorted.vcf -L /path/to/Agilent_exome_v2_44mb_1kGenomeB37_baits.interval_list -o /path/to/BTXXX_WES_MuTect2.vcf

Answers

  • SheilaSheila Broad InstituteMember, Broadie, Moderator admin

    @david.knoff
    Hi,

    Can you please post the original bam files and bamout files for the two samples in that region?

    Thanks,
    Sheila

  • david.knoffdavid.knoff Boston, MAMember
    edited March 2016

    Hi @Sheila ,

    When I try to attach these files, it's telling me that the file type is not allowed. Can you message me an email to send these files to? I attached the IGV images for the original bam and bamout files around this variant.

    One interesting thing I'm seeing is that when I isolated the interval to just surrounding this variant, the variant gets called correctly with MuTect2. However, when I used a targeted interval list of about 400 genes (which includes IDH1), the variant did not get called during my first run for some reason. I am repeating the MuTect2 targeted panel analysis to see if I receive the same results. I am also running a whole exome analysis with MuTect2.

    Thanks,
    David

  • SheilaSheila Broad InstituteMember, Broadie, Moderator admin

    @david.knoff
    Hi David,

    The screenshots are perfect. No need to send over the bam files.

    So, when you generated the bamout file you posted here, you ran on a small interval surrounding the site and it gets called? However, when you run on a different interval surrounding the site, it does not get called? Can you please post the record from the VCF where it does get called? I am wondering if the quality score is low. The downsampling is done based on the intervals, so it is entirely possible that with different intervals, you may get different calls. However, these calls tend to be "edge cases" where the quality score is not very high and depending on the reads chosen, the call may be made or not.

    -Sheila

  • david.knoffdavid.knoff Boston, MAMember

    @Sheila
    Hi Sheila,

    I attached 3 files from my IDH1 focused interval analysis. BTXXX_IDH1.txt is the original vcf output from MuTect2 (I changed it to .txt because the website wasn't allowing me to upload .vcf files). BTXXX_IDH1_oncotator.txt is the vcf output from oncotator (input was BTXXX_IDH1.vcf). BTXXX_IDH1_oncotator_table.txt is the output from the GATK VariantsToTable function (input was BTXXX_IDH1_oncotator.vcf).

    I did another run with the OncoPanelv3 interval list (provided by CCGD at DFCI) and the IDH1 mutation did not appear once again. This is WES data so I will try my WES interval list to see if this makes a difference.

    Thanks,
    David

  • SheilaSheila Broad InstituteMember, Broadie, Moderator admin

    @david.knoff
    Hi David,

    Let us know what you find with the exome intervals. Can you confirm that the base qualities and mapping qualities are good at that site? Also, please post an IGV screenshot of the surrounding region as well. I want to know if the region is messy and if there is a potential larger event going on there.

    Thanks,
    Sheila

Sign In or Register to comment.